Parameter Identification at Large Scale through Solution of the Inverse Problem

ahmeda m salem

Abstract

The fundamental benefit of groundwater modeling is to predict the response of the aquifer to the proposed man-mad excitation (e.g., pumping artificial recharge). Inverse analysis is frequently used for large-scale real-world problems, where the number of unknown parameters that will fully represent the heterogeneity becomes very large and the inverse analysis solution becomes ill-posed, to overcome this problem, in the present paper the study domain is divided into zones connected to the hydraulic gradient using different methods of parameterization. Zoning characterized with similar hydraulic gradient was considered having the same values of transmissivity, accordingly deferent zoning schemes of parameter distribution were obtained then the optimization process was conducted. The determination of the optimum dimension of the parameter distribution was achieved utilizing the trade-off between the objective function and the error of uncertainty.Keywords: Parameter identification, transmissivity, zonation, inverse analysis